Doubly robust identification for causal panel data models

Summary We study identification and estimation of causal effects in settings with panel data. Traditionally, researchers follow model-based strategies relying on assumptions governing the relation between potential outcomes observed unobserved confounders. focus a different, complementary approach to identification, where are made about connection treatment assignment Such common cross-section settings, but rarely used introduce different sets that two paths develop double robust approach. propose methods build these strategies.